ce-clean-gone-branches

Pass

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using branch names retrieved from the local git environment. This presents a risk of command injection if a branch name contains shell metacharacters (e.g., ;, $(), or backticks). Evidence: The instruction in Step 3 of SKILL.md uses the $branch variable directly in commands like git worktree remove --force "$worktree_path" and git branch -D "$branch".
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted local data (branch names) and uses it to drive agent actions without validation.
  • Ingestion points: Branch names are ingested from the output of git branch -vv via the scripts/clean-gone script as specified in SKILL.md.
  • Boundary markers: Absent; branch names are displayed to the user and passed to tools without delimiters or instructions to ignore embedded commands.
  • Capability inventory: The skill has the capability to delete branches and remove worktrees via shell commands in SKILL.md.
  • Sanitization: Absent; neither the discovery script nor the workflow instructions perform escaping or validation on the branch names before they are used in commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 23, 2026, 02:55 PM